Steam Linux Usage Hit 0.80% During November

[skeevy420]

I think it just means that a higher percentage of Linux users have more CPUs.. aka <0.00 Windows have 32 cores, but that still could be 1000s of users. This also tracks because there are some CPU values at 0 only on the Windows report such as 28.

>>> x = 1 + 28.07 + 1.75 + 58.45 + .01 + 9.13 + 1.45 + .06 + .04 + .01 + .04 + .01
>>> print(x)
100.02000000000004

That's the numbers for Windows in November added up. Close enough to 100% including rounding errors. But, yeah, 0.00% probably means negligible, but they technically exist and the - probably means no users reported.

Maybe Steam needs to use thousandths due to the sheer number of users they have so low number of users doesn't appear to be no users.

Based on their numbers, there's around 200 of those 32 core Linux people. One of them has to post here.

Or 8. I'm not sure if I should use 40K or 125M.

[Morbis55]

I dumped my Windows dual boot installation completely this year. The only thing i am missing is Freesync support.
But else, Steam Play changed everything. I was too lazy to use Wine or PlayonLinux. Steam Play just works and is 0 effort.

So no more need for Windows for most Games without Linux support. And the rest you may not even want to play anyway.
Screw you DRM protection software. Bugger off!

[sdack]

It's exciting to see, but it means very little at this time. Too much isn't working yet, or just not flawlessly, and it's still early in the beta phase (I see it more as being in an alpha state).

It all depends on how much effort Valve is going to put into Steam Play / Proton. Knowing how the gaming industry often likes to rush their crappy code out of the door am I not excited, but sceptic over the future.

I believe the market share could go a lot higher if Valve plays their cards right. A lot of people are drawn to it now, but a lot of it doesn't work or just not flawlessly. Yet everyone tries to provide feedback or even fixes for their favourite games. Valve needs to pick up on it, but I fear they might just rush the whole thing out of beta too soon.

[MJaoune]

I think it just means that a higher percentage of Linux users have more CPUs.. aka <0.00 Windows have 32 cores, but that still could be 1000s of users. This also tracks because there are some CPU values at 0 only on the Windows report such as 28.

Linux
4 cpus 47.52% -1.13%
6 cpus 8.41% +0.49%
8 cpus 7.25% +0.07%
16 cpus 0.45% +0.03%
20 cpus 0.03% +0.03%
22 cpus 0.01% +0.01%
32 cpus 0.02% +0.02%

Windows
4 cpus 58.45% +1.41%
6 cpus 9.13% +1.09%
8 cpus 1.45% +0.29%
16 cpus 0.04% 0.00%
20 cpus 0.00% 0.00%
22 cpus 0.00% 0.00%
28 cpus 0.00% 0.00%
32 cpus 0.00% 0.00%

It seems some system administrators might be turning data centers into netcafes. Don't be surprised if 8.8.8.8 joins your game someday.

[Brisse]

An interesting trend looking at the numbers is how AMD is starting to pickup on the CPU front and take numbers from Intel.

If you look at Linux users only then AMD GPU's are clearly on the rise as well. I guess the open source drivers are convincing a lot of Linux users to go for AMD GPU's.

I also noticed that some Linux Steam, and only Linux Steam, user is using a 32 core CPU to play games. I just chuckled and figured it almost has to be a Phoronix user.

It's probably Michael while benchmarking the Threadripper 2990WX

[Weasel]

I think it would really be in Valve's interest to try getting PUBG working

Ditch anti-cheat and it will work.

[mphuZ]

Wait for statistics to update. The current figures are clearly incorrect.

[xiando]

That being said, I've used steam with Windows for maybe five years, and got a single survey. Used steam with linux for about a year, and got two surveys. So I don't believe the steam surveys are biased.

I got no surveys for a year and then a survey two months in a row. It's probably random and not biased.

The bigger question is why they are using a "survey" system at all, it really seems quite stupid. Mine was just something like "Do you want to be in the survey" and I clicked next or whatever it said and that was it. It seems obvious to me that they could just have a configuration option under Privacy settings which says "submit anonymous hardware survey data monthly" or similar. It seems kind of trivial to do. I'd be fine with it as long as it's opt-in.

[Dedale]

It is too early to draw conclusions IMHO.

[atmartens]

I got no surveys for a year and then a survey two months in a row. It's probably random and not biased.
The bigger question is why they are using a "survey" system at all ... they could just have a configuration option ... I'd be fine with it as long as it's opt-in.

And then it's no longer random, because you can't guarantee that the people who opt in are reflective of the general population (the same problem might also be true with the current system). What if Linux users were twice as likely to opt in? Better would be to just update the frequency with which people are surveyed, given that it seems to be rare. However, if the surveys are too frequent, then people are more likely to skip it (imagine if it happened every day!)